This paper presents a method for choosing the most appropriate contact model from several models, by comparing their efficiency for predicting hysteretic behaviour in different applications. The first part describes an application for a belt tensioner used for automotive engines. The second part concerns an academic application and focuses on a simple case of vibration in the presence of dry friction. The hysteresis loops obtained at contact locations are usually used to reproduce the nonlinear forces transmitted along the contact surface as accurately as possible. Although such techniques can be used when the contact zone is easily accessible, this is not generally possible for real, complex structures. Consequently, different techniques are developed and validated

In this paper we describe the modeling of the hysteretic behavior of belt tensioners. An initial experimental device is composed only of the tensioner by using forcing frequencies, preloads and deflection amplitudes. It permits the identification of the parameters of the restoring force model used. Comparison of the measured and predicted force-deflection loops of the tensioner subjected to large deflections permits preliminary validation of the model. The second experimental device consists of a belt-tensioner system. Its non-linear modeling includes the above hysteretic model and the belt’s longitudinal characteristics. Validation of the belt-tensioner model is completed by comparing the measured and predicted belt tension. Finally, it is shown by using a parametric investigation and phase-plane portrait that the response of the belt-tensioner system increases with the frequency and the amplitude of the excitation.